Re: meta content-language

On Aug 21, 2008, at 12:55 PM, Henri Sivonen wrote:
> On Aug 21, 2008, at 22:46, Roy T. Fielding wrote:
>> On Aug 20, 2008, at 11:57 PM, Henri Sivonen wrote:
>>> Is there software that acts on the HTTP header or meta Content- 
>>> Language in the sense "This is a document for people who read both
>>> English and French"? What does software acting on that meaning do?
>>
>> That is the definition for the header field Content-Language,  
>> which is
>> what is referred to by META http-equiv, and yes it is used by some
>> content management systems as a means for authors to define metadata
>> that will be returned by HTTP in a response and/or used in content
>> negotiation to choose the most appropriate representation.
>
> How is it used in content negotiation so that software *acts on* it  
> (as opposed to acting on something else and outputting the outcome  
> in HTTP Content-Language)?

It is extracted from the documents within a given directory to create
a mapping of URIs to representation alternatives, which is then used
during the response process.

>> WGN is
>> one of the HTTP servers that works that way by default, though I  
>> don't
>> know if anyone uses it now.  Apache can do that via modules.
>
> Do WGN and Apache use Content-Language as the input for a decision?

WGN does, IIRC.  Apache does if you enable the module for deriving
meta maps from content (or from separate .var or .meta files when
tied to the output of a simple perl script).  CERN had .meta dirs
for the same purpose, also capable of being populated with a cron.

>> In any case, all of the http-equiv attributes are defined by HTTP.
>> That is its definition in HTML.
>
> It's not the definition in HTML5 as drafted.

Then add it to the things that are formal objections to HTML5.

....Roy